The basis of multiple-input/multiple-output (MIMO) operation is to provide wireless devices with multiple radio interfaces to allow the devices to send data on different channels at the same time in order to achieve greater transmit/receive data rates and with greater reliability. In MIMO systems, a transmitter sends multiple streams of encoded data packets to a receiver by multiple transmit antennas. The streams may be spatially and time encoded and converted into multiple RF signals. The signals are transmitted to the receiver on multiple channels between multiple transmit antennas at the transmitter and multiple receive antennas at the receiver. When the receiver receives the signal vectors from the multiple receive antennas, the receiver decodes the received signal vectors into the original information.
A spatially multiplexed MIMO system that uses multiple transmit and receive antennas not only transmits data between the corresponding transmit and receive antennas but also between adjacent antennas. Thus, data is received in the form of a MIMO channel matrix. Linear algebra techniques such as singular value decomposition (SVD) or matrix inversion may be required to decouple the channel matrix in the spatial domain and recover the transmitted data. The transmitter typical requires some knowledge of the channel state to effectively transmit the streams. One approach for estimating the channel state is to use channel reciprocity, which is generally based on the theory that if a link operates on the same frequency band in both directions, an impulse response of the channel observed between any two antennas may be the same regardless of the direction.
In a MIMO system having a m transmit antennas and n receive antennas, an (n×m) time varying matrix H is typically denoted as the channel matrix representing the physical propagation channel, where each column represents a channel gain from each transmit antenna of the transmitter to n receive antennas of the receiver.
The channel by which the transmitter transmits the data stream to the receiver is referred to as the forward channel, and may be represented as a channel matrix Hf. The channel from the receiver to the transmitter is referred to as backward channel, and may be represented as channel matrix Hb. Channel reciprocity means that a forward channel and a backward channel are equivalent. Mathematically, channel reciprocity can be defined as:HbT=Hf where T is matrix transpose operation.
A forward channel matrix is a transposed version of the backward channel matrix. For example, the forward channel from transmit antenna 1 to receive antenna 2 is the same as the backward channel from receive antenna 2 to transmit antenna 1.
MIMO performance has been improved through the use of beamforming techniques. Beamforming allows multi-antenna radios to communicate multiple streams of information across a multipath channel such that all streams use the same radio spectrum but do not interfere. Beamforming takes advantage of interference to change the directionality of an antenna array. When transmitting in beamforming, the transmitter is the beamformer and the receiver is the beamformee. The phase and relative amplitude of a signal of beamformer is controlled in order to shape the transmitted beam pattern narrower, such that the energy is transmitted in a particular direction of the beamformee, in contrast to an omni-directional beam pattern that transmits energy in every direction. When used in a WLAN or cellular environment, beamforming can result in increased received signal power and reduced interference power at the receiver/mobile station.
Several types of beamforming are known, such as beamforming with full channel knowledge and beamforming with no channel knowledge. Beamforming with full channel knowledge can be achieved via two different techniques. One technique for determining full channel knowledge is for beamformer to transmit known training sequences from beamformer transmit antennas to receive antennas of the beamformee to enable the beamformee to estimate channel state information and determine the full channel matrix Hf. Then the beamformee feeds back the forward channel Hf to the beamformer.
Another technique for determining full channel knowledge may be referred to as implicit beamforming. Implicit beamforming calls for the beamformee to “sound the backward channel,” wherein the beamformee sends a known signal to the beamformer. The beamformer then estimates the channel state information for Hb and infers Hf based on channel reciprocity.
Once the beamformer determines full channel knowledge of the forward channel, i.e., the full channel matrix Hf, the beamformer can perform beamforming. In a downlink situation where the beamformer and the beamformee know Hf, they can employ Singular Value Decomposition (SVD) to use input and output singular vectors of Hf to spatially multiplex and demultiplex the transmitted and received vectors to form multiple spatial filters, called beams, with their antenna arrays. In other words, the beams are “steered” in the direction of the receiver. The result of this mux/demux operation is that information symbols in x are communicated through the channel matrix in parallel and without inter-symbol interference. The received symbols are the transmitted symbols scaled by a corresponding singular value, S, but may be corrupted by background noise.
Beamforming may also be performed with no channel knowledge. In beamforming with no channel knowledge, the beamformer randomly generates the steering vector without knowledge of the forward channel to the beamformee. For example, the beamformer may randomly generate a steering vector such that at time 0, a signal is transmitted in a North direction; at time 1, a signal is transmitted in an East direction; at time 2, a signal is transmitted in a South direction; and at time 3, a signal is transmitted in a West direction. Beamformees that receive a strong signal may send a feedback signal reporting that the signal was received and beamformees that received a weak signal may send a feedback signal reporting that the received signal was weak. The beamformer may then decide to which reporting beamformees to allocate the forward channel. Beamforming with no channel knowledge is effective when there are many beamformees associated with a given a station because the beamformer beamforms to an arbitrary direction, and in most cases, the beamformees will be spread over a whole coverage area in all directions, particularly in cellular systems.
Although beamforming with full channel knowledge and beamforming with no channel knowledge are effective techniques, in some situations, only partial channel knowledge exists. In some MIMO systems, the number of transmit chains in the beamformee can differ from the number of receive chains. For example, in many conventional WiMAX systems, beamformees may have two receive chains, but only one transmit chain, while in WiFi systems, beamformees may have three receive chains and two transmit chains. Typically, the number of transmit chains is smaller than the number of receive chains.
In implicit beamforming based on the beamformee sounding the backward channel, it is assumed that the beamformee sends the known signal to the beamformer on all transmit antennas in order for the beamformer to determine the forward channel. Sometimes, however, the beamformee may not send the known signal to the beamformer using all available transmit antennas. That is, the beamformee may sound only from a subset of available transmit antennas. In this case, only channels from a subset of beamformee transmit antennas may be known to the beamformer. So equivalently, only a partial channel matrix, i.e., a subset of columns of the backward channel Hb, is known. And through channel reciprocity, only a subset of rows of the forward channel Hf will be known to the beamformer. Thus, in this situation, beamforming needs to be done based on the partial channel knowledge, i.e., only a subset of rows of the forward channel Hf.